Assessing accuracy and specificity of faecal source library for microbial source-tracking, using SourceTracker as case study
Bibliographic record
Abstract
Motivation: Understanding the quality of the source library prior to undertaking library-dependent microbial source-tracking (MST) is an essential, but often overlooked, primary analysis step. Results: We propose an assessment approach to validate the quality of amplicon-derived faecal source libraries. This approach was demonstrated on a faecal source library consisting of 16S rRNA paired-end amplicon sequences, obtained from various animal types in Victoria, Australia. First, a leave-one-out (LOO) analysis was performed to assess the accuracy of source category groupings by identifying the number of samples incorrectly assigned to a different source category (i.e. animal type). Following a quality control procedure to decide retaining/removing/grouping incorrectly assigned samples, we then assessed if the sample sizes for each source type were sufficient to properly characterize the source fingerprints. Results from LOO demonstrated 15.5% of samples were incorrectly assigned, with high error rates in birds and wallabies within our source library. Increasing the sample size improved source identification accuracy. However, accuracy eventually plateaued in a source-specific manner. Importantly, this highlights the importance of conducting thorough assessments to understand the quality and limitations of the source library prior to library-dependent MST applications. Availability and implementation: QIIME2 is available via https://qiime2.org/; SourceTracker v2.0.1 is available via https://github.com/caporaso-lab/sourcetracker2; Pipeline for LOO is available via https://github.com/MonashOWL/Bioinformatics-IlluminaMGI/tree/main/16S/LOO; Pipeline for sample size assessment is available via https://github.com/MonashOWL/Bioinformatics-IlluminaMGI/tree/main/16S/Source%20variability.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.016 | 0.037 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.003 | 0.002 |
| Open science | 0.001 | 0.003 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".